Parking slot detection on fisheye images plays a crucial role in vehicle planning, and has recently become a rapidly advancing domain in autonomous driving. Previous work struggle with practicality due to their lacks of large-scale parking slot dataset and inflexible representations of parking slot. To alleviate this, we release ParkScape, the first large-scale fisheye dataset for parking slot detection. It has 72,000 corner-based parking slots that covering abundant diversities in scenes, density, and location. Moreover, we build a fisheye parking slot detector based on a dense single-stage object detection framework. With the detector constructed, a novel Dynamic Feature Fusion (DFF) module is proposed to mitigate the fisheye distortions through integrating context features. Our proposed DFF module can be easily plugged into existing keypoint detectors to improve the performance. Extensive experiments on ParkScape validate the superiority of our fisheye parking slot detector, achieving an impressive mean Average Precision ( AP ) of 47%, while operating at a remarkable speed of 54 frames per second (FPS). The dataset and code will be available at: https://github.com/Vipermdl/ParkScape.
Image restoration under adverse weather conditions poses a challenging task. Previous research efforts have predominantly focused on eliminating rain and fog phenomena from images. However, snow, being another common atmospheric occurrence, also significantly impacts advanced computer vision tasks such as object detection and semantic segmentation. Recently, there has been a surge of methods specifically targeting snow removal, with the majority employing visual Transformers as the backbone network to enhance restoration effectiveness. Nevertheless, due to the quadratic computations required by Transformers to model long-range dependencies, this significantly escalates the time and space consumption of deep learning models. To address this issue, this paper proposes an efficient snow removal Transformer with a global windowing network (SGNet). This method forgoes the localized windowing strategy of previous visual Transformers, opting instead to partition the image into multiple low-resolution subimages containing global information using wavelet sampling, thereby ensuring higher performance while reducing computational overhead. Extensive experimentation demonstrates that our approach achieves outstanding performance across a wide range of benchmark datasets and can rival methods employing CNNs in terms of computational cost.